Note: Descriptions are shown in the official language in which they were submitted.
CA 02356693 2001-06-26
WO 00/39565 PCTII1'S99/30835
AN iN-LINE PROCESS FOR MONITORING BINDER
DOSAGE AND DISTRIBUTION ON A SURFACE
AND APPARATUS USEFUL THEREFOR
BACKGROUND OF THE INVENTION
The present invention relates to a process for monitoring binder,
particularly isocyanate-based binders, dosage and distribution on a
surface by ultraviolet fluorescence and to apparatus useful therefor.
Various types of binders have been used to produce engineered!
composite materials such as oriented strand board. Suitable binders
include phenol formaldehyde resins and isocyanates, particularly
polymeric diphenylmethane diisocyanate {"PMDI"). In producing such
engineered composites, the binder is generally applied to a material such
as wood fibers, wood strands, wood flakes or some other lignocellulose-
based material. Ideally, the amount of binder applied ("dosage"} would be
sufficient to cover 100% of the surface of 100% of the wood fibers, wood
strands, etc. {"distribution"). In most commercial processes, an excess of
binder is used to ensure sufficient distribution. Longer than necessary
mixing times may also be used to ensure that the binder is sufficiently
distributed so that weak spots in the composite material due to insuffcient
adhesion do not occur. This use of excess binder and extended mixing
times significantly increases the cost of producing engineered composite
materials.
It would therefore be advantageous to develop a method for
determining binder dosage and distribution during the composite
production process with sufficient accuracy that use of excess binder and
extended mixing times are unrsecessary.
Spectroscopic methods for making such determinations have been
investigated by those seeking to improve the production of composite
materials. Solid NMR characterization of the bonding of composite
CA 02356693 2001-06-26
WO 00/39565 PCT/tUS99130835
-2-
materials was studied by Frazier and Wendler and the results were
presented in "15N CP/MAS NMR analysis of pMDI bonded ceNulose
composites" presented at the 48th Annual Meeting of the Forest Produc,~ts
Society, Portland, Maine, ,tune 2fi-29, 1994. Sun et al attempted to
correlate fluorescence intensity changes with FT1R spectra generated by
monitoring the disappearance of the isocyanate group during the reaction
which occurs in the commercial production process. (See, e.g., Sun et al,
Institute of Materials Science, Storrs Report TR-38-~JNR, Connecticut
University (1994).)
UV absarption and fluorescence spectroscopy are also techniques
which have been evaluated for their usefulness in monitoring urethane-
forming reactions. However, until now, methods utilizing such UV
spectroscopic techniques were not capable of providing real time,
macroscopic imaging of the composite material as it was being produced.
For example, F. Kamke's work reported in "Wood Based
Composites Program Annual Report" (June 1, 1994 - May 31, 1995) was
a microscopic study of UV fluorescence imaging of polymeric MDI resin
distribution on wood strands. Kamke states that because polymeric MD!
fluorescence is very weak, a very intense UV source (specifically, a 100
watt mercury vapor lamp) and signal averaging of many video frames to
reduce noise level were necessary. Although signal averaging to reduce
noise Level works well for stationary samples, it is not very useful when the
material being evaluated is moving on a conveyor belt and the video
image is constantly changing. Microscopic evaluation of a material is also
impractical for monitoring a commercial production process because of the
great potential for variation between samples. The Kamke method would
not therefore be practical for monitoring a commercial process for the
production of a composite material.
Yu et al report a technique in which naphthylene diisocyanate is
used as a molecular sensor to monitor cure reactions in a polyurethane in
U.S. Patent 4,885,254. Yu et al correlate the fluorescence intensity and
CA 02356693 2001-06-26
WO 00/39565 PCT/US99I30$35
-3-
overall extent of reaction between 1,5-naphthyl diisocyanate and n-
butanol. This correlation was established by identifying the various
species present during the urethane-forming reaction using HPLC that was
confirmed by IR spectra. The UV-visible absorption spectrum and
fluorescence spectrum for each of these species were then generated.
Shifts in the UV-visible spectrum were observed as the naphthyl
diisocyanate reacted to form the monourethane and diurethane. The
fraction of each species present at a given time was determined by linear
regression analysis. The extent of the reaction was calculated from U1/
spectra! analysis. A correlation between the experimentally determined
fluorescence intensity at 357 nanometers and the calculated overall extent
of reaction derived from UV spectral analysis was made.
U.S. Patent 5,100,802 discloses a method for measuring the rate
and extent of cure of a resin system in which a fluorescent dye 'ss added to
the system being polymerized.
U.S. Patent 4,922,113 discloses a method for monitoring a
coating's weight, uniformity and surface defects in which a UV-escer that
absorbs radiant energy is included in the coating composition. The radiant
energy emitted by the coating at the same wavelength as energy emitted
by the UV-escer can be detected and correlated to pre-established
standards.
U.S. Patent 4,651,011 discloses a method for determining the
extent of cure of a polymer. In this method, the degree of free space
rotation of a fluorospore added to the polymer system is determined by
fluorescent measurement of the fluorospore.
To date, however, no method for determining binder dosing,
particularly isocyanate-based binder dosing and distribution during actual
production of composite materials without adding some type of "marker"
such as a dye, fluorospore or UV-escer has been developed.
CA 02356693 2001-06-26
WO 00/39565 PCT/U~S99130835
,.4_
SUMMARY OF THE INVENTION
ft is an object of the present invention to provide an apparatus for'
determining binder dosage and distribution during the production of a
composite material.
It is also an object of the present invention to provide an apparatus
for determining binder dosage and distribution while the composite-forming
material is moving, for example, on a conveyor belt.
It is another object of the present invention to provide an apparatus
for accurately determining binder dosage and distribution during
production of a composite from a composite forming material that does not
include a marker material such as a UV-escer, fluorospore or dye.
It is a further object of the present invention to provide an in-line
_- - - . . - - -monitoring-process which is capable of detecting binder dosage
and
distribution with a high degree of accuracy even though the material being
monitored is not stationary.
These and other objects of the present invention which will be
apparent to those skilled in the art are accomplished by exposing the
composite-forming material to long wave ultraviolet radiation to excite the
binder molecules sufficiently to induce fluorescence and emission of
visible radiation white passing the composite-forming material through the
field of view of a video camera equipped with a suitable filter. The video
camera is positioned to receive the visible fluorescence emitted by the
binder molecules. The video camera is connected to a device capable of
correlating the visible fluorescence collected to pre-established standards
such as a computer programmed with an appropriate database.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 is a schematic representation of an apparatus suitable for
determining binder distribution by UV fluorescence in which the results are
color, enhanced before being displayed.
CA 02356693 2001-06-26
WO 00/39565 PCT/~'JS99/30835
_5_
Figure 2 is a reproduction of chromatographic analyses of the
polymeric MD1 (i.e., polyphenyi polymethylene polyisocyanate) in the
binder composition used to coat the wood strand images shown in Figures
3-7. These chromatographs were generated using a Hewlett Packard size
exclusion chromatograph having dual UV-absorption and UV-fluorescence
detectors.
Figure 3 is a false color representation of the UV fluorescent image
of wood strands coated with 0% polymeric MDI.
Figure 4 is a false color representation of the UV fluorescent image
of wood strands coated with 2% polymeric MDI.
Figure 5 is a false color representation of the UV fluorescent image
of wood strands coated with 4% polymeric MDI.
Figure 6 is a false color representation of the UV fluorescent image
' of wood strands coated with 6% polymeric MDI.
Figure 7 is a false color representation of the UV fluorescent image
of wood strands coated with 8% polymeric MDI.
Figure 8 is a set of histograms of the monochrome images used to
generate the false color images in Figures 3-7 in which the number of
image elements or pixel population at each level of brightness for the gray
scale between 0 and 255 is plotted along the Y axis with the brightness or
gray scale level plotted along the X axis.
Figure 9 is a linear regression of the predicted binder dosage based
on average brightness of the monochrome images used to generate the
false color images in Figures 3-7 to actual binder dosage.
Figure 10 is a reproduction of the transmission spectra of the filters
used in the filter "sandwich" through which fluorescent waves are passed
before those waves are passed through the lens and video camera.
Figure 11 is a 2-dimensional contour map of the UV fluorescence of
the polymeric MD# wood binder and wood rot.
CA 02356693 2001-06-26
WO 00/39565 PCT/US99/30835
-s-
Figure 12 is a monochrome representation of the UV fluorescent
image of wood strands coated with 5% of polymeric MDl generated by
using a blue band pass filter to eliminate wood rot fluorescence in the
image.
Figure 13 is a true color image of wood strands coated with 5%
polymeric MDI obtained using a color camera.
Figure 14 is the image of Figure 13 in which the wood rot is masked
out.
DETAILED DESCRIPTION OF THE PRESENT INVENTION
The present invention is directed to an in-line process for monitoring
binder, particularly isocyanate-based binder, dosage and distribution
during the production of composite materials and to apparatus useful
therefor.
The apparatus of the present invention is composed of a long wave
ultraviolet radiation source, a filter or a set of filters, a lens, a video
camera
and a means for correlating the ultraviolet fluorescence data collected from
a sample to the dosage and distribution of binder in that sample. A typical
arrangement of these components of the apparatus of the present
invention is shown in Figure 1.
As shown in Figure 1, long wave ultraviolet radiation 1 (i.e.,
electromagnetic radiation or light having wave lengths between 300 and
400 nm) emitted from lamps which emit Tong wave ultraviolet waves 2 is
directed to the sample 3 (i.e., the composite-forming material or substrate)
in a manner such that only light from the UV source, lamps 2, illuminates
the sample 3 and that the flux density of UV radiation 1 is constant across
the surface of sample 3. The binder molecules present in sample 3
absorb UV radiation 1 emitted by lamps 2 and fluoresce or emit visible
radiation 4 (light having characteristic wavelengths between 400 and 700
nm}.
Visible radiation 4 emitted by the fluorescing binder in sample 3 is
passed through filter 5 and tens 6 before being imaged by video camera 7.
CA 02356693 2001-06-26
WO OOI39565 PCT/US99/30835
-7-
Filter 5 as shown in Figure 1 is a "sandwich" composed of three individual
filters. The first filter 5a (i.e., the filter closest to sample 3) is a UV
blocking
filter. The second filter 5b (or flter next closest to sample 3) is a long
pass
fitter having a cutoff in the blue spectrum. The third filter 5c (or fitter
nearest video camera 7) is a visible band-pass filter.
Video camera 7 may be a monochrome or color video camera. If a
monochrome video camera is used, the wavelength of the visible band-
pass filter 5c in filter 5 is chosen so that only those wavelengths specific
to
the binder fluorescence will pass through. In practice, a color video
camera has been found to be more versatile, particularly in cases where
interference from another material that may fluoresce (e.g., waxes, mold
release agents, preservatives and wood rot) may be present. When a
color video camera is used as camera 7, the visible band-pass filter 5c in
filter 5 may be eliminated and the tong-pass filter 5b can be used to block
any wavelengths shorter than 435 nm. Both color and monochrome video
cameras are generally equipped with an internal near infrared blocking
filter (not shown). A near infrared blocking ~Iter is necessary to eliminate
any near infrared emissions of UV source 2.
In Figure 1, filter 5 is mounted directly onto lens 6. However, it is
not necessary that filter 5 and lens fi be physically connected or attached.
The relative positions of filter 5 and lens fi may also be reversed. It is
sufficient to arrange filter 5 and lens fi so that only visible radiation 4
that
has passed through filter 5 and lens fi is imaged onto camera 7.
UV source 2, filter 5, lens 6, and camera '7 are positioned within
housing 8 in the apparatus shown in Figure 1. As shown, housing 8 is
open at the bottom to allow the passage of UV radiation 1 from UV source
2 to sample 3, and visible radiation 4 from sample 3 to filter 5. Housing 8
also prevents outside light from illuminating sample 3 and or passing
through filter 5. Housing 8 is a preferred but not an essential feature of the
apparatus of the present invention.
CA 02356693 2001-06-26
WO 00139565 PCTIIdS99130835
_$-
The images captured by video camera 7 are relayed via video
cables) 9 to image processing hardware 10. Image processing hardware
is typically a computer equipped with a high speed video frame capture
board (not shown) and software capable of image analysis. When a
5 monochrome video camera 7 is used, the image analysis software
enhances the contrast between the fluorescing binder and the non-
fluorescing background of sample 3, calculates the binder dosage based
on the average image intensity, and displays the image as a false color
image. The false color image is generated by converting the gray scale of
10 the monochrome image to a color palette as shown in Figures 3-7 with the
color palette used shown as a strip along the bottom of the image going
from low to high left or right
When a color video camera 7 is used, the image analysis software
analyzes each of the red, green and blue images separately in order to
discriminate the binder from any other interfering fluorescent material,
calculates the binder dosage and displays the binder distribution.
The determined binder dosage and distribution levels may then be
displayed on the monitor 11, printed or converted to an analog signal for
output to other instrumentation. The results may be in the form of an
image of the sample in the actual fluorescing color or a false color image
with a pallet reflecting binder dosage.
Under actual production conditions, sample 3 would typically be on
a conveyor belt that is moving at a rate of from about 5 to about 30 feet
per minute. The field of view of video camera 7 may be adjusted with
zoom lens fi. The #ield of view is generally set to cover the maximum area
that is evenly illuminated by UV source 2, preferably an area 2 feet long
and 1.5 feet wide, directly below camera 7. At 30 feet per minute, for
example, it takes about 3 seconds for the composite material 3 imaged by
video camera 7 to move completely out the field of view of camera 7. This
time frame is more than sufficient for the image analysis software to
analyze the image for dosage and distribution of the binder and to display
CA 02356693 2001-06-26
WO OOI39565 PCT/US99/30835
_g_
the results of this analysis. Continuous, real-time monitoring of a
composite production process is therefore achieved with the apparatus of
the present invention.
The long wave ultraviolet radiation source 2 may be any device
which emits long wave ultraviolet radiation, i.e., radiation having
wavelengths between 300 and 410 manometers, preferably between 300
and 400 manometers. Examples of suitable ultraviolet radiation sources
include lamps with mercury vapor bulbs having long wave ultraviolet filters
and lamps having long wave ultraviolet fluorescent bulbs such as Genera!
Electric ultraviolet bulbs such as bulbs F40T12IBLB (commercially
available from GE Lighting of Cleveland, Ohio).
White it is possible to use only one ultraviolet lamp as the UV
source in the apparatus of the present invention, it is preferable that more
than one lamp be included in the apparatus. It is particularly preferred that
from 8 to 20, most preferably about 14, lamps be included in the
monitoring apparatus of the present invention to ensure even illumination
of the sample surface being monitored.
In the embodiment of the present invention illustrated in Figure 1,
fourteen lamps with 40 watt fluorescent UV long wave bulbs each of which
is 48 inches Tong (GE F40T12IBLB) are spaced 1.75 inches from center.
The transmission spectrum of the UV filter glass of the GE F40T12/BLB
bulb is shown in Figure 10 as Curve A. The emission spectrum of the
phosphorus used in the GE bulbs is centered at 360 nm. The emission
spectrum of the GE UV fluorescent bulbs follows Curve A in the spectrum
reproduced as Figure 10 between 300 and 400 nm.
The composite-forming material 3 may be made up of any of the
materials, particularly lignoceliulosic materials, known to be useful for the
production of composite materials to which a binder, preferably an
isocyanate-based binder, has been applied. Examples of suitable
composite-forming materials 'seclude: wood, wood fibers, wood bark, cork,
bagasse straw, flax, bamboo, esparto, rice husks, sisal fibers and coconut
CA 02356693 2001-06-26
WO 00/39565 PCT/tTS99/30835
_10_ _
fibers. Wood strands, shavings and chips used in the production of
engineered lumber known as oriented strand board .("OSB") are
particularly preferred. The strands, chips or shavings may vary in size
from 0.03" x 0.18" x 2.0" to 0.12" x 2.0" x 28.0". The preferred size ranges
from 0:03" x 1.0" x 16" to 0.12" x 2.0" x 24" and the most preferred range is
from 0.03" x 0.75" x 3.0" to 0.12" x 2.0" x 24". These lignocellulosic
materials may have a moisture content of from about 0.5 to about 30% by
weight, based on total weight of material, preferably from about 3 to about
8% by weight when used in~the production of a composite material.
Any of the binder compositions having a fluorescing component that
are known to be useful for the production of composite articles may be
used to produce composite articles in accordance with the present
invention. Such known binders include isocyanates. The preferred binder
- compositions are polyphenylene poiymethylene polyisocyanates
("polymeric MDI") which contain higher molecular weight oiigomers (i.e..,
ofigomers having more than four rings} that have an inherent fluorescence
and thus do not require the addition of a fluorescing agent to the binder
composition. The binder should have a minimum of 35% by weight of
oligmers with more then four rings.
Figures 2A and 2B are reproductions of the chromatographic
analyses of the polymeric MDI used in the Examples. These
chromatographs were generated using a Hewlett Packard size exclusion
chromatograph (Model 1090M) with dual UV absorption and UV
fluorescence detectors. The UV fluorescence peak in Figure 2A that
eluted between 26 and 36 minutes (maximum at 30.8 minutes) is
associated with the highest molecular weight oligomers of Figure 2B
eluting out on the UV absorption detector. The UV fluorescence peak
essentially disappears upon the appearance of the 4-ring, 3-ring and
monomer peaks at 35.4, 36.7 and 39.0 minutes respectively in Figure 2B.
Monitoring isocyanate binder fluorescence thus appears to be a viable
method for monitoring a commercial wood composite production process.
CA 02356693 2001-06-26
WO 00139565 PCT/US99130835
-11-
Ideally, the binder would be used in an amount sufficient to coat
100% of the surface area of the material to be bonded. If even distribution
of the binder is achieved, the amount of binder necessary to achieve
approximately 100% coverage is minimized. However, in commercial
production processes, the binder dosage may be as much as 50 to 60%
greater than the amount theoretically necessary to achieve 100%
coverage in order to ensure complete coverage.
Filter 5 shown in Figure 1 is a multi-layer "sandwich" arrangement
of filters. The first layer 5a (nearest to sample 3) is a UV blocking filter
that
protects the other filters and video camera CCD (Charge Coupled Device)
imaging detector from UV damage. Any commercially available UV
blocking filter that blocks UV radiation wavelengths shorter than 380 to
400 nm (preferably shorter than 400 nm) may be used as the blocking filter
5a.' Examples of suitable blocking filters include filters made with the
polycarbonate plastic having a UV stabilizer that are commercially
available under the names Makrolon (Bayer Corporation) and Lexan (~E
Plastics). A 0.125 inch thick sheet of Makrolon plastic filter is particularly
preferred. Curve B in Figure 10 is a transmission spectrum of optical
grade polycarbonate with the UV blocking cutoff at 400 nm.
The second filter layer 5b of filter 5 is a long-pass filter that blocks
the UV and short wavelength blue between 400 and 455 nm (preferably all
wavelengths shorter than 455 nm). Any of the commercially available
long-pass filters with a sharp cutoff at 455 nm may be used as fitter 5b in
filter 5. The 455 nm long-pass filter F45,063 which is commercially
available from Edmund Scientific, Barrington, N,i) is particularly preferred.
Curve C in Figure 10 is a transmission spectrum of the Edmund Scientific
455 nm long-pass filter F45,063.
The third filter layer 5c of filter 5 is a visible band-pass filter selected
to increase the contrast between the binder and the composite material
and eliminate any interference from other fluorescing components. For
example, if the binder composition is based on polymeric MDI and camera
CA 02356693 2001-06-26
W4 00/39565 PCT/US99I30835
-12-
7 is a monochrome video camera, the preferred visible band-pass filter 5c
of filter 5 is a blue filter. Any of the commercially available blue band-pass
filters with a band-pass between 400 and 530 nm would be suitable. A
visible band-pass filter that has been found to be particularly useful in the
apparatus of the present invention is Edmund Scientific's Night Bfue
transparent acrylic filter from filter assortment F39418. Curve D in Figure
is a transmission spectrum of this Night Blue transparent acrylic fitter.
Filter 5 need not be a sandwich structure. It is possible to have a
custom optical interference filter made that would have all of the
10 characteristics of filter 5a, 5b, and 5c. However the cost would be
considerably higher unless a large number of filters were made at one
time. The illustrated filter sandwich cost less than $100. A custom
interference filter may cost in excess of $500.
Any of the commercially available lenses may be used in the
practice of the present invention. The lens does not have to be a zoom
lens, however, a zoom lens makes it easy to adjust the field of view to
cover the maximum area that is evenly illuminated by the UV source 2. A
particularly preferred lens 6 is Edmund Scientific's 8 to 48 mm zoom lens
F53152.
Any of the commercially available video cameras may be used in
the practice of the present invention as video camera 7. This camera may
be either a monochrome or a color camera. Examples of suitable video
cameras include the Sony Model XC-75 monochrome camera and Cohu
Model 2222-2340 color camera. Each of these cameras is available from
Edmund Scientific, Barrington, New Jersey.
Filter 5, lens 6 and camera ? are positioned with respect to sample
3 in a manner such that the field of view of camera 7 is adjustable with
lens 6. Video camera 'T is set to cover the maximum area that is evenly
illuminated by UV source 2, It is preferred that there be a distance of
about 48 inches between the surface of sample 3 and lens 6 and that the
field of view be an area of about 2 feet in length and about 1.5 feet in width
CA 02356693 2001-06-26
WO 00/39565 PCTIUS99/30835
-13-
directly below camera 7. Filter 5, lens 6 and camera 7 are positioned with
respect to each other in a manner such that a!i light falling on the CCD
(Charge Coupled Device) image detector of camera 7 will have passed
through filter 5 and Lens 5.
Housing 8 is a supporting structure that is helpful for maintaining
the alignment of UV source 2, filter 5, lens 6 and camera 7 and shielding
the composite-forming material 3 in the field of view of camera 7 from
external light. Housing 8 will usually be open at its bottom to allow
passage of UV radiation from UV source 2 to the composite-forming
material 3 and to allow passage of the fluorescent light back to filter 5,
lens
6 and camera 7.
Any of the commercially available computers with a high-speed
video frame capture board may be used as image processing hardware 10
in the apparatus of the present invention. Computers which have been
faund to be particularly advantageous are those having a 233 MHz or
faster Intel Pentium Processor with video frame capture board such at that
which is available under the name "Bandit" from Coreco, lnc., Quebec,
Canada.
Any of the commercially available image processing software
programs that can be installed on computer hardware l0 may be used in
the apparatus of the present invention. One software program that has.
been found to be particularly advantageous in the practice of the present
invention is the Wit Visual Programming Software that is available from
Logical Vision, Quebec, Canada.
Any of the commercially available monitors may be used in the
practice of the present invention as monitor 11. it is preferred, however,
that any monitor employed be capable of displaying images and/or spectra
in color.
The material to be monitored in accordance with the present
invention (i.e., the substrate) is generally prepared by mixing or blending a
iignocellulose-based material with the binder, depositing the composite-
CA 02356693 2001-06-26
WO 00/39565 PCTIU~99/30835
-14-
forming mixture or blend into a mold or a form of some sort on a conveyor
belt, passing that composite-forming material under housing 8 to monitor
dosage and distribution of the binder, and passing that composite forming
material through a heated press to cure the binder composition. The press
is typically maintained at a temperature of from about 50 to about
210°C,
preferably from about 150 to about 200°C to ensure cure of the binder.
Figures 3-7 show the UV fluorescence images of wood strands that
have been coated with polymeric MDl at levels of 0%, 2%, 4%, 6%, and
8% by weight binder, based on the weight of the wood strands, using a 5
foot diameter lab blender (available from Coil Inddstries, Vancouver,
Canada). The images in these Figures 3-7are false color representations
of the monochrome gray scale. Wood rot fluoresces very brightly and
appears red and red orange in color. Wood strands with no binder are
dark blue in color and strands coated with binder appear fight blue to
green to yellow in color as the binder dosage increases. Examination of
the central two-thirds of the images where the CCD (Charge Coupled
Device) image detector illumination is constant, shows that there is a very
uneven distribution of binder on the strands. If the binder shown in Figure
5 at 4% dosage had been distributed evenly over al! of the strands, 4%
dosage would be sufficient to produce the composite article. Figure 6
shows that even at a 6°l° dosage level same strands are not
adequately
coated. Figure 7 shows that at 8% dosage all strands are covered with
binder but many strands are over-dosed and the excess binder is wasted.
The ability to monitor binder dosage and distribution achieved with the
apparatus of the present invention will make it possible to optimize the
production process and achieve significant savings due to the use of less
binder.
Having thus described our invention, the following Examples are
given as being illustrative thereof. All parts and percentages given in
these Examples are parts by weight and percentages by weight, unless
othennrise indicated.
CA 02356693 2001-06-26
WO OOI39S65 PCT/US99/30835
-15-
EXAMPLES
EXAMPLE 9
An apparatus corresponding substantially to that shown in Figure 1
was used to analyze long strands of aspen wood to which polymeric MDI
(Mondur 541 which is commercially available from Bayer Corporation) was
applied at levels of 0%, 2%, 4%, 6% and 8% to calibrate the device for
dosage.
The UV source 2 was made up of eight lamps having 4 foot 40 watt
bulbs (F40T12/BLB bulbs available from GE Lighting, Cleveland, Ohioj,
and four lamps having 2 foot 20 watt bulbs (F20T121BLB bulbs available
from GE Lighting, Cleveland, Ohio). The lamps were arranged
symmetrically around filter 5, lens fi and video camera 7 as shown in
- -- -- -- Figure 1 to evenly illuminate the aspen strands which were 48
inches from
the UV tamps 2.
Composite-forming material 3 was composed of aspen wood
strands ranging in size from 0.08" x 1.0" x 6" to 0.12" x 1.5" x 12" and
having a moisture content of 6% by weight. 0%, 2%, 4%, 6% or 8% by
weight of polymeric MDI which is commercially available from Bayer
Gorporation under the name Mondur 541 was applied to these strands.
Filter 5 was a filter sandwich made up of two layers. The first layer
was a polycarbonate UV blocking filter made up of 0.125" thick Makrolon
polycarbonate with UV stabilizers (commercially available from Shefsield
Plastics, Shefsield, MA). The second filter layer was composed of the 455
nm long-pass filter F45,063 (available from Edmund Scientific; Barrington,
N. ,!.).
Lens 6 was the lens F39,087 (available from Edmund Scientific,
Barrington, N. J.) having f11.3-1fi and 8.5 mm focal length.
Video camera 7 was a Sony model XC-75 monochrome camera
(available from Edmund Scientific). A 75-ohm coaxial cable 9 connected
video camera 7 to image processing hardware 10.
CA 02356693 2001-06-26
WO 00/39565 PCTlUS99/30835
-16-
Image processing hardware 10 was a 233 MHz Pentium processor
based computer with 128 Megs of RAM. The computer was a model
DELL OptipIe:GXA sold by Dell Computer Corporation. Bandit Video
Frame Capture Board sold by Coreco Inc, Quebec, Canada was used to
digitize the video signal from camera 7.
The image processing software algorithms were programmed using
the Wit Visual Programming Software (available from Logical Vision,
Quebec, Canada).
The aspen wood strands were dosed with polymeric MDI in a six
foot diameter lab blender (Coil Industries, Vancouver, Canada). The
strands were dosed with binder in 2% increments and then passed under
the monitoring apparatus of the present invention. The aspen wood
strands were covered with a cardboard shield to prevent exposure to the
UV radiation 'until the image processing software was ready to capture the
video image. Exposure to UV radiation for periods of longer than 95 to 30
seconds will bleach the composite material and reduce the overall
fluorescence enough to affect the calibration. Actual exposure to the UV
radiation to capture an image takes only 0.1 second. After the image was
captured, the wood strands were placed back in the blender to be dosed
with another 2% of binder. This process was repeated until images at 0%,
2%, 4%, 6% and 8% binder were obtained.
The monochrome images were then processed using the Wit
software by adjusting the contrast equally on each image and converting
to a false color pallet to represent the gray scale. The false color images
are shown in Figures 3-7. Because the filters used will allow alt visible
light having wavelengths between 455-700 nm, there is no discrimination
between the binder and the other fluorescing materials such as wood rot.
However, the intensity of the yellow fluorescence of the wood rot is
considerably greater than the blue green fluorescence of the polymeric
MDI binder. When the monochrome images are converted to false color,
the high intensify wood rot is shown as orange and red and the binder is
CA 02356693 2001-06-26
WO 00!39565 PCT/ITS99/30835
-17-
shown as light blue to green to yellow depending upon the binder dosage.
The 0% resin image of Figure 3 shows some light blue in the center of the
image which is the result of the weak natural deep blue fluorescence of the
wood under long wave UV and higher tight throughput in the center of the
image, a characteristic of the f/1.3-16, 8.5 mm focal length lens used as
lens 6.
The image processing software was then used to calculate the
histogram for each of the images. A histogram is a mathematical
representation of the image in which a vector is generated by assigning a
vector element for each of the 256 brightness levels of the gray scale. The
value of each vector element is the total number of pixels in the image at
that intensity.
Figure 8 is a graphical representation of the histograms generated
from the images in Figures 3-7. In Figure 8, the brightness or gray level
between 0 and 255 arbitrary unites is plotted on the x-axis. The number
of pixels or pixel population is plotted on the y-axis. Curve A in Figure 8
represents the histogram for the wood strands dosed with 0% binder.
Curve B in Figure 8 represents the histogram of the wood strands dosed
with 2% binder. Curve C in Figure 8 represents the histogram of the wood
strands dosed with 4% binder. Curve D in Figure 8 represents the
histogram of the wood strands dosed with fi% binder. Curve E in Figure 8
represents the histogram of the wood strands dosed with 8% binder. The
wood rot shows up as the positive step in the histograms at the upper end
of the x-axis.
The average image intensity was calculated from the histograms by
taking the cross product sum of the vector (i.e., the sum of each vector
element value times that vector element's number [a number from 0 to
255]) and dividing by the total number of pixels. This calculation of
average image intensity is different from the calculation for the area under
the curve of the histogram. The area under the curve of the histogram
would always be equal to the number of pixels (i.e., 307,200 for a fi40 x
CA 02356693 2001-06-26
WO 00139565 PCT/CTS99I30835
-18-
480 pixel image). The cross product sum weights each element of the
vector with the intensity of that element. The average image intensity can
then be correlated with the binder dosage using linear regression. The
results of such linear regression are shown in Figure 9.
EXAMPLE 2
The apparatus used in Example 1 was modified to eliminate the
interference of wood rot fluorescence when a monochrome camera is
employed.
The number of lamps used as the UV source was increased. A
total of fourteen lamps having 4 foot 40 watt F40T121BLB bulbs {GE
Lighting, Cleveland, Ohio) were used as UV source 2. A third layer was
added to filter sandwich 5. This layer was a visible band-pass filter
selected to increase.the.-contrast between the binder and the lignocellulose
material and to eliminate any interference from the fluorescing wood rot.
The filter used was Edmund Scientific's Night Blue transparent acrylic filter
from filter assortment F39418. Curve D in Figure 10 is a transmission
spectrum of the Night Blue filter. Lens 6 was replaced with an 8 to 48 mm
zoom tens (lens F53152 available from Edmund Scientific, Barrington,
N.J.).
Figure 11 is a 2-dimensional contour map of the UV-fluorescence of
the poiymeric MDI (Mondur 541 available from Bayer Corporation) and a
yellow fluorescing wood rot extract made by soaking wood rot in
tetrahydrofuran solvent for 6 hours. This UV contour map was generated
using an Hitachi Model F4500 Fluorescence Specrophotometer UV
fluorescence spectrometer. In Figure 11, the binder fluorescence is seen
as the thick line contours and the wood rot fluorescence is seen as the thin
fine contours. The x-axis is the excitation wavelength, and the y-axis is the
emission wavelength. The UV source lamps 2 have an emission spectrum
following that of the lamp UV glass filter transmission spectrum shown as
Curve A in Figure 10 between 300 and 400 nm.
CA 02356693 2001-06-26
WO 00/39565 PCT/UrS99/30$3~5
_19_
Given the emission spectrum of the UV source lamps 2 (Curve A in
Figure 10), the transmission spectrum of the 455 nm long-pass filter
(Curve C in Figure 10), and the Night Blue transparent acrylic filter
spectrum (Curve D in Figure 10), it can be seen from Figure 11 that the
majority of the wood rot fluorescence could be eliminated from the camera
image.
Images were taken of wood strands dosed with 5% binder using the
above-described modified apparatus. Figure 12 is a monochrome image
taken with this modified apparatus. It is evident from Figure 12 that the
fluorescence of the wood rot was eliminated from the image when
compared to the true color image of Figure 13. Figure 12 shows the
uneven distribution of binder over the wood strands. A vertical strand in
the center of the image. was partially masked by_other strands,in the bun_d~r~
. .~
dosing process.
EXAMPLE 3
The apparatus used in Example 2 was further modified to allow the
quantitative and qualitative analysis of multiple fluorescing materials in the
composite-forming material. This was accomplished by eliminating the
third layer of the filter sandwich (filter 5c), i.e., the Night Blue
transparent
acrylic visible band-pass ~Iter. A Cohu Model 2222-2340 color camera
(available from Edmund Scientific) was substituted for the camera used in
Example 2.
Images of wood strands dosed with 5% polymeric MDl were then
taken with this modifred apparatus: Figure 13 is a full color image from
camera 7 in which wood rot is represented by the yellow green
_ fluorescence, the isocyanate binder is represented by the light blue
fluorescence and the un-dosed wood is represented by the dark blue
fluorescence. This image was then processed by splitting the full color
image into red, green and blue images. The yellow green fluorescing
wood rot had a component in the red and in the green image but virtually
none in the blue image. The light blue-fluorescing isocyanate binder had a
CA 02356693 2001-06-26
WO 00/39565 PCTIUS99/308:15
-20-
component in the blue and in the green images but none in the red image.
A mask was generated from the monochrome red image using all pixels
above a level of 85 on a gray scale of 0-255 to eliminate all wood rot pixels
from the green and blue images. The full color image with the masked
wood rot is reproduced as Figure 14. The average image intensity of the
unmasked pixels for the blue and green images could then be calculated
in the same manner as was used in Example 1. This average image
intensity was then correlated to binder dosage. The amount of wood rot
could be correlated to the average image intensity of the red image.
Additional fluorescing components could be analyzed by examining their
red, green and blue images and applying the appropriate image masking
techniques or multivariate analysis to each corresponding pixel in the three
images.
Although the invention has been described in detail in the foregoing
for the purpose of illustration, it is to be understood that detail is solely
for
that purpose and that variations can be made therein by those skilled in
the art without departing from the spirit and scope of the invention except
as it may be limited by the claims.
